GPU Overclocking: Stability

TechyInAZ · Apr 7, 2016

I guess I'm confused as to how a GPU is stable. I currently was able to get my GTX 750 Ti FTW up to 1455mhz (including GPU boost 2.0) and a 1500mhz memory clock at a good stability.

But, it was only stable in fullscreen mode. Even with the card at 1445mhz core clock and 1450mhz memory clock, I can't get it stable with a game running in windowed mode AND having a youtube video running in the background.

Why is this?

thor220 · Apr 7, 2016

A couple of things come to mind. First, did you check for stability using something like FurMark for a few hours to assess stability? Have you tried the same scenario with an un-overclocked video card? You might also want to try switching browsers to see if it's just the one as well.

TechyInAZ · Apr 7, 2016

Thank you, and no, I'll admit I didn't stress test the card for even 30 minutes. I'll do that when I have time.

Yes, I've been doing that with the stock clock on my video card for a year now, it's not had a single issue.

Ryan_78 · Apr 7, 2016

Maybe just run ungine heaven benchmark and check temps artifacting. If nothing is bad, you're fine. Remember you should t just pull the sliders right, add like 30 or 40 MHz at a time and check.

thor220 · Apr 7, 2016

TechyInAZ :

Thanks for the response. Yeah, I would go ahead and stress test. I'd say that 3 hours is the minimum to try to simulate a decent gaming session.

Is there any pattern to the games crashing other than being in windowed mode? Like, for example, does it happen in every game that's windowed?

Any chance you can get some data on your GPU usage, clock speeds, mem speed, ect before and up to a crash?

TechyInAZ · Apr 8, 2016

thor220 :

Yeah I can get data on GPU usage and stuff like that, however it'll have to wait until tomorrow.

The data on the GPU voltages, core clock, TDPs etc. are fine before and after every crash (mostly after every crash). GPU boost 2.0 is always cranked up at 1455mhz without ANY dips what so ever. Power percent is always drifting between 50-75%. Memory is the same also. Since the ACX cooler is completely overkill for this card, temps are always 58C or lower on the lowest fan settings possible.

Typically what happens is when I'm in game and running the YT vid at the same time, It'll run perfectly for about 5-10 minutes. Then soon after that, I will noticed very minor artifacting then the GPU driver completely crashes.

So far I've gone from a offset of 110+ on GPU core clock and 200+ on vram clock, to 100+ GPU core clock and 150+ memory clock and get the same issues.

Should I try 90+ on the core clock?

Oh and BTW, my memory is Elpida which I've heard is the worst on the market, so that could have something to do with it, maybe?

thor220 · Apr 8, 2016

TechyInAZ :

thor220 :

Yeah I can get data on GPU usage and stuff like that, however it'll have to wait until tomorrow.

The data on the GPU voltages, core clock, TDPs etc. are fine before and after every crash (mostly after every crash). GPU boost 2.0 is always cranked up at 1455mhz without ANY dips what so ever. Power percent is always drifting between 50-75%. Memory is the same also. Since the ACX cooler is completely overkill for this card, temps are always 58C or lower on the lowest fan settings possible.

Typically what happens is when I'm in game and running the YT vid at the same time, It'll run perfectly for about 5-10 minutes. Then soon after that, I will noticed very minor artifacting then the GPU driver completely crashes.

So far I've gone from a offset of 110+ on GPU core clock and 200+ on vram clock, to 100+ GPU core clock and 150+ memory clock and get the same issues.

Should I try 90+ on the core clock?

Oh and BTW, my memory is Elpida which I've heard is the worst on the market, so that could have something to do with it, maybe?

Elpida is kinda the bargain RAM, definitely not know for it's overclocking. Since the memory is working fine on stock clocks that probably means it's the culprit here. You should try lowering just the memory clock to see if that fixes the issue. I know I've had AMD cards with Elpida RAM and they exhibit similar issues when overclocked. Things like screen flickering, artifacts ect. In certain cases the ram would run funky on even a small overclock.

TechyInAZ · Apr 8, 2016

Alright thanks. It's a bummer my card didn't come with Samsung vram like most other people say they have. But oh well.

thor220 · Apr 8, 2016

TechyInAZ :

Yeah, samsung RAM is pretty great. I really don't get why they mix RAM types even on the same model card. If they are going to give you cheaper RAM they should at least give you a discount.

TechyInAZ · Apr 9, 2016

Agreed. But I guess that's not how things work around here.

Quick update: I kept the core clock at +100, and lowered the vram to +140 on the mem clock. This time I ran Furmark and another video at the same time, i did get like one artifact, but with how insanely high furmark stresses the GPU and with it trying to play a vid at the same time, it's fine. And i got no instability. This is the first time I've used furmark, so much better than the heaven benchmark.

I'll try and duplicate the YT vid + playing war thunder in windowed mode again to see if it glitches out. If so, then I wonder if it's a nvidia driver issue.

thor220 · Apr 9, 2016

TechyInAZ :

Looks promising so far. If the issue is only appearing in very specific cases and not in things like Benchmarks and other games it could very well be a driver issue.

Search

GPU Overclocking: Stability

TechyInAZ

Titan

thor220

thor220

Splendid

TechyInAZ

Titan

Ryan_78

Honorable

thor220

Splendid

TechyInAZ

Titan

thor220

Splendid

TechyInAZ

Titan

thor220

Splendid

TechyInAZ

Titan

thor220

Splendid

TRENDING THREADS

Latest posts

Moderators online

Share this page